Scholarlink

<p>Cellular neural networks, inspired in part by the biological retina, offer a potential route to massively parallel analogue computing. However, the hardware implementation of such systems remains challenging. Here we report memristor-based cellular neural networks for image and video processing applications. We develop a Python-based digital twin for network simulations and as a graphical user interface for controlling the fabricated hardware. Simulations using the digital twin illustrate the network’s capabilities in image processing and in solving partial differential equations. We build hardware through the tape-out of a transistor-based network and the fabrication of a circuit board with multilevel non-volatile memristors as the synapses. We show that the hardware can be used to run image processing tasks including edge and horizontal line detection.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Memristive cellular neural networks for fast in-pixel computing

  • 期刊论文

  • 发表日期：

    2026年1月19日

查看全文
  • Vignesh Ravichandran,
  • Yi Huang,
  • Bryce Flannery,
  • Tergel Molom-Ochir,
  • Tina Maurer,
  • Shiva Asapu,
  • Ali Abdel-Maksoud,
  • Nia Heermance,
  • Remy Yoo,
  • Joshua Tackie,
  • Wuyu Zhao,
  • Yunzhi Ling,
  • Alex Guo,
  • J. Joshua Yang,
  • Qiangfei Xia

摘要

Cellular neural networks, inspired in part by the biological retina, offer a potential route to massively parallel analogue computing. However, the hardware implementation of such systems remains challenging. Here we report memristor-based cellular neural networks for image and video processing applications. We develop a Python-based digital twin for network simulations and as a graphical user interface for controlling the fabricated hardware. Simulations using the digital twin illustrate the network’s capabilities in image processing and in solving partial differential equations. We build hardware through the tape-out of a transistor-based network and the fabrication of a circuit board with multilevel non-volatile memristors as the synapses. We show that the hardware can be used to run image processing tasks including edge and horizontal line detection.